Anthropic vs Pentagon: Political and Legal Tensions

A Legal Conflict in California

A California court recently made a notable decision by temporarily suspending the Pentagon's actions against Anthropic. The judge prevented the government from classifying the company as a supply chain risk while ordering agencies to stop using its artificial intelligence technologies. This ruling comes amid a conflict that has persisted for a month, and the government has seven days to appeal this decision. Meanwhile, Anthropic has initiated another legal proceeding regarding this designation, which remains pending.

The Stakes of the Confrontation

From the outset, the stakes in this case were clear: it concerns the government's ability to sanction a company that does not meet its expectations. Anthropic has received support from several influential figures, including former AI policy advisors from the Trump administration. However, Judge Rita Lin emphasized that this contractual dispute should never have escalated to such intensity. According to her, the government ignored established procedures for managing such conflicts, exacerbating tensions with contradictory public statements.

A Controversial Use of AI

Court documents reveal that the government used Anthropic's Claude technology for much of 2025 without raising any complaints. The company, which positions itself as a safety-focused AI player, secured defense contracts despite growing disagreements. Defense employees, using Claude via Palantir, were required to accept specific conditions prohibiting mass surveillance and lethal autonomous warfare. Tensions emerged when the government sought to establish a direct contract with Anthropic.

Incendiary Public Statements

The situation took a more confrontational turn when disagreements were made public, appearing more motivated by a desire to punish than by a simple breach of contract. President Trump posted a message on Truth Social, labeling Anthropic as "left-wing lunatics" and ordering federal agencies to stop using its services. Secretary of Defense Pete Hegseth quickly followed suit, stating his intention to designate Anthropic as a supply chain risk.

Contested Government Actions

The judge found that the actions required for such a designation had not been properly executed by Hegseth. Letters to congressional committees indicated that less drastic measures had been considered but deemed impossible, without providing details. Furthermore, the government claimed that Anthropic could activate a "kill switch," but had to admit in court that it had no evidence of this.

A Violation of Constitutional Rights?

The public statements led the judge to conclude that Anthropic had valid reasons to complain about a violation of its First Amendment rights. The government appeared to want to publicly punish the company for its ideological beliefs and reluctance to compromise them. Designating Anthropic as a supply chain risk would equate to labeling it a "saboteur," an accusation for which the judge found insufficient evidence.

A Judicial Decision with Significant Consequences

The judge issued an order suspending the designation, preventing the Pentagon from enforcing it. Dean Ball, a former AI policy advisor under Trump, described this decision as a "devastating judgment" for the government, emphasizing that Anthropic is likely to prevail on most of the government's arguments. Although the government is considering an appeal, another similar case is underway in Washington.

Implications for the Future

Court documents show a disconnect between the public statements of officials and legal procedures. The Pentagon and the White House knew that this approach would lead to a legal battle. Anthropic had already announced its intention to contest the designation even before it was officially filed. This situation distracted senior officials during the early days of the war in Iran, when they could have managed the fallout with Anthropic more straightforwardly.

An Uncertain Future for Anthropic

Even if Anthropic emerges victorious, the government has other means to exclude it from public contracts. Defense companies, eager to maintain good relations with the Pentagon, might hesitate to collaborate with Anthropic, even without an official risk designation. Charlie Bullock, a researcher at the Institute for Law and AI, points out that the government can exert pressure without breaking the law, depending on its level of investment in punishing Anthropic.

A Cultural War over AI

The administration appears to be dedicating significant resources to this cultural battle over AI. Claude, Anthropic's technology, is deemed crucial enough that President Trump stated the Pentagon would need six months to stop using it. The White House demands political alignment from AI companies, but the case against Anthropic reveals the limits of its current power.